Chinese scientists have taken a significant step toward understanding and controlling the behavior of complex quantum systems, demonstrating that quantum computers can track — and even regulate — processes that are effectively impossible for classical computers to calculate.

Using a 78-qubit superconducting quantum processor known as "Chuang-tzu 2.0", researchers from the Chinese Academy of Sciences'Institute of Physics and the Peking University observed and controlled a phenomenon called prethermalization, a temporary and orderly phase that appears before a quantum system descends into full chaos.

Their findings were published online on Wednesday in the journal Nature.

Quantum systems behave very differently from everyday objects. When many quantum particles interact, their collective behavior becomes extraordinarily complex, and information spreads rapidly through the system. Over time, this leads to thermalization — a process in which energy and information are evenly distributed and the system loses memory of its original state.

For quantum computing, thermalization is a major problem. Once a system goes through this process, fragile quantum information is effectively lost, making reliable computation impossible.

What surprised scientists is that thermalization does not always happen smoothly or immediately.

"In certain conditions, the system pauses," said Fan Heng, a corresponding author of the study and a researcher at the Institute of Physics. "It enters a stable intermediate stage where disorder is delayed and information is partially preserved."

That stage is known as prethermalization.

Researchers liken it to heating a block of ice. Even when heat is continuously applied, the temperature remains at 0 C for a time as the ice melts. Energy is channeled into changing the structure rather than raising the temperature. In a similar way, a driven quantum system can absorb energy without becoming fully chaotic.

In the experiment, scientists deliberately "pushed" the quantum processor using carefully designed pulses of energy. Instead of applying simple, repeating signals, they used a method called "Random Multipolar Driving". The technique introduces structured randomness into the driving pattern, based on mathematical sequences that are neither fully periodic nor completely random.

By adjusting the pattern and timing of these pulses, the team was able to control how long the quantum system remained in the prethermalized state before rapidly descending into chaos.

"This allows us to tune the rhythm of thermalization," Fan said. "We can slow it down or speed it up."

During the prethermal phase, measurements showed that quantum information remained relatively intact and the growth of disorder was suppressed. Once the phase ended, however, quantum entanglement — a key feature of quantum mechanics — increased and spread rapidly across the system, making the system too complex for classical computers to simulate accurately.

Scientists noted that the results could influence future research into quantum simulation, quantum control, and related topics.

The team plans to develop larger quantum chips with more flexible architectures, aiming to explore even more complex quantum behavior and to demonstrate what they call "verifiable, practical quantum advantage", the point where quantum machines don't just do things faster, but solve specific, useful problems that were previously impossible.